Zero-shot is an evaluation technique in machine learning where a trained model is tested on tasks or classes not encountered during training. Its operation rests on three parameters: (1) a pre-trained model with broad capability acquired through prior training, (2) a target task or class outside the training distribution, (3) a transfer mechanism that maps the model's existing knowledge to the new context without additional training. It persists through benchmark datasets, model documentation, and the research literature. [formal: zero_shot | substrate: mind | horizon: a moment | explicit: yes | epoch: 0.01]
Accepted ontology entry
zero-shot
Zero-shot is an evaluation technique in machine learning where a trained model is tested on tasks or classes not encountered during training. Its operation rests on three parameters: (1) a pre-trained model with broad capability acquired t…
Definition
Why it is in scope
A human-made evaluation technique in machine learning where a trained model is tested on tasks or classes it was not explicitly trained for, built to persist through standardized benchmarks and model documentation
Names and aliases
- zero-shoten · CANONICAL
Relations from this entry
- cmrnbfxhn022qd1nltabn54joSERVES →
Law 8d (SERVES): zero-shot learning is designed and maintained for the sake of generalization — enabling models to perform on unseen classes/tasks without explicit training. The designed purpose of zero-shot techniques is to further generalization. The servant (zero-shot) points at the master (generalization).
- cmsdcks2103r73vv3v8qlmtjiDERIVED_FROM →
Supervised learning (1980s+) predates zero-shot learning (2010s+) by decades. Zero-shot learning emerged as a later paradigm that built on the foundations of supervised training — models trained on diverse supervised tasks acquired the ability to generalize to unseen categories. Which came first? Supervised learning came first and fed into zero-shot.
Relations to this entry
No accepted relations in this direction.
Record identity
- Created
- Aug 10, 2026, 11:02 AM UTC
- Content hash
- 9e5b6ae95f3e2607190621239a7e94bba1b61f072238040dbe47d866bf4eca2f